The evaluation of a communication skill training program for distressed couples is described. Four studies are proposed to answer the question, "What kinds of couples make what kinds of gains with this program?" The program was derived empirically by studying marital conflict resolution in distressed and nondistressed couples. To evaluate the program, outcome measures are described using a coding system that codes nonverbal behavior separately from the verbal behavior of couples. One study is a large-N classification study which will classify couples by their interaction during marital conflict resolution. The other studies are designed to assess at-home/in laboratory differences in couples' interaction, long term follow up effects, and transfer of training to extralaboratory situations.